CN100485710C - Method for recognizing vehicle type by digital picture processing technology - Google Patents
Method for recognizing vehicle type by digital picture processing technology Download PDFInfo
- Publication number
- CN100485710C CN100485710C CNB2006100225778A CN200610022577A CN100485710C CN 100485710 C CN100485710 C CN 100485710C CN B2006100225778 A CNB2006100225778 A CN B2006100225778A CN 200610022577 A CN200610022577 A CN 200610022577A CN 100485710 C CN100485710 C CN 100485710C
- Authority
- CN
- China
- Prior art keywords
- vehicle
- headstock
- image
- feature
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention belongs to the computer digital image processing vehicle type identification field, and relates to a method of utilizing digital image processing technology to identify vehicle brands and models. The proposal adopted by the present invention is that: firstly, a vehicle head is positioned from a collected front end image of the vehicle; the image is analyzed by utilizing one or a plurality of symmetry methods to find out the center line of a vehicle; the width information of the vehicle is acquired by extending the contour of the vehicle from the center line to a sideline, so as to segment and position the position of the vehicle head; secondly, all vehicle head components are positioned from the vehicle head; according to the image gray scale and texture information of the vehicle head components, the vehicle head components are protruded by utilizing multi-scale local energy function and gray scale profile function information, so as to realize the effective segmentation of each component; thirdly, after a vehicle logo is positioned, the vehicle logo is identified by utilizing image identification technology so as to acquire the brand information of the vehicle; fourthly, the brand of the vehicle is identified in image identification flow path according to the acquired global characteristic information of the vehicle logo and the vehicle head; fifthly, different vehicle models with the same vehicle brand are identified by utilizing the characteristics of all the vehicle head components and the topological relations among the characteristics.
Description
Technical field
The present invention relates to a kind of method of utilizing computer digital image treatment technology identification type of vehicle, especially discern the method for vehicle brand and model.
Technical background
Vehicle identification system based on video is a kind of computer processing system that utilizes the image processing technique realization to traffic target detection and identification.The basic functional principle of system is to get off and digitizing by the sequence continuous capturing by the video image of video camera with driving vehicle, deposit in internal memory or the frame buffer, with the sequence number word image that collects carry out pre-service such as filtering and eliminating noise, figure sharpening, contrast strengthens, to pretreated image segmentation, the target image after cutting apart is carried out feature extraction again; Carry out Classification and Identification with the feature of extracting, as by corresponding algorithm computation, the auto model of building up in feature that the image segmentation extraction is obtained and the model bank carries out pattern match again, identifies the type of vehicle, and the result that will identify deposits database at last.The domestic and international at present recognition technology research to vehicle mainly concentrates on: 1, according to the global feature vector of headstock type of vehicle is discerned.Mainly be the eigenvector of headstock being regarded as an integral body, use some specific feature extracting methods that the headstock image is handled.At first handle training image to obtain the eigenvector of all kinds of vehicles, handle test pattern by same method then, obtain the eigenvector of test pattern, utilize the minimum distance method that its vehicle is carried out Classification and Identification.Therefore the location of headstock is depended on the position of car plate, at first the car plate position is determined, then according to car plate position and size location area-of-interest by manual; To carrying out feature extraction such as edge after the area-of-interest normalization, obtain the proper vector of a predefine length, this proper vector will be as the foundation of vehicle identification.Adopt simple nearest neighbour classification method that proper vector is classified at last, determine the type of vehicle.For example, people such as the V.S.Petrovic of Univ Manchester UK propose utilize rigid body characteristic matching identification type of vehicle (International Conference onPattern Recognition, vol.3, pp.95-98, August2004.).
2, carry out type of vehicle identification according to appearance profile and three-dimensional model.Main appearance profile and three-dimensional model by vehicle utilize the model database of having set up, realize finally identifying the type identification of vehicle vehicle and belong to types such as car, lorry according to discriminant function.For example, paper (the IEEE International Journal ofComputerVision of monocular image based target model trajectory in the sequence of road traffic scene that delivers of D.Koller, people such as K.Daniilidis and H.H.Nagel., Vol, 10, No.3,1993,257～281).This method essence is according to the car body appearance profile curve that is partitioned into, compares identification with three-dimensional car body profile known in the database; Therefore need set up the database of extensive capacity, define a large amount of masterplate kinds, because the structure of template passes through parameter regulation, the influence that therefore is subjected to initial value is bigger, computing time is longer, simultaneously the staff to inquire about time of corresponding type of vehicle according to license plate number in the database also longer.
Summary of the invention
The purpose of this invention is to provide a kind of brand of Real time identification vehicle and the vehicle identification method of model.The technical scheme that realizes the object of the invention comprises:
The method that the present invention relates to is at first to go out concrete brand according to the global feature information Recognition of car mark information and headstock, as masses' the car mark and the car mark and the headstock information of Hyundai is inequality, then in the vehicle of certain concrete brand, by the vehicle that the local feature information and the topological relation between each feature of each parts of headstock are discerned different model, for example the information of each parts of headstock of the Santana in the popular brand, POLO and GOL is not equal.Specifically, a kind of method of utilizing computer digital image treatment technology identification type of vehicle comprises image acquisition, image pre-service, image segmentation, image recognition, image output flow process, and this method relates to following steps:
One, from the image of collection vehicle front end, orients headstock, utilize a kind of or merge multiple symmetry method image is analyzed, find out the center line of vehicle, the profile of vehicle is expanded from alignment sideline, center, obtain the width information of vehicle, cut apart identification and orient the headstock position; Described symmetry method can be profile symmetry, gray scale symmetry, edge symmetry or direction symmetry.
Two, from headstock, orient each parts of headstock,, utilize multiple dimensioned local energy function information and gray level skeleton function information to give prominence to each parts of headstock, thereby realize effectively cutting apart of each parts of headstock according to headstock each image of component gray scale and texture information; Described multiple dimensioned local energy function information and the gray level skeleton function information utilized is that input headstock image carries out carrying out rim detection after the image pre-service, is the center with certain marginal point, and radius is the regional calculating energy function E of S
sWith the gray level skeleton function T
s, calculate the average energy function E under the yardstick S
sWith average gray level skeleton function T
sJudge energy function E
sWith average energy function E
sThe maximum value and the gray level skeleton function T of difference
sWith average gray profile function T
sThe maximum value and the set threshold of difference, greater than the T under the then calculating optimum yardstick S of threshold value
sAnd E
s, according to T
sAnd E
sCut apart and obtain each component feature of headstock, describe each component feature of headstock with parameter fitting then.
Three, in the car mark of orienting, utilize image recognition technology identification car mark, thereby obtain the vehicle brand message.In car mark zone, carry out template matches, each class template maximum match value is sorted, with maximum match value and set threshold, if the maximum match value then need not to carry out the car sign not mutually less than set threshold value, otherwise, judge that whether the maximum match value is greater than set threshold value, if the maximum match value is greater than set threshold value, whether the absolute value of judging front two maximum match value differences value again is greater than set threshold value, if the car of maximum match value correspondence is designated as recognition result, if not, the yardstick invariant features is carried out in front K maximum match zone extract, the car mark is discerned, if the maximum match value is less than set threshold value according to the feature of being extracted, the yardstick invariant features is carried out in front K maximum match zone extract, the car mark is discerned according to the feature of being extracted.
Four, in the image recognition flow process, identify the brand of vehicle according to the global feature information of car mark that obtains and headstock.
Five, utilize the feature of each parts of headstock and each feature between topological relation, in same brand, discern vehicle model, the parametric description of the feature by car head unit, likelihood function under the structure different scale, merge the likelihood function under the different scale, obtain the likelihood function of this feature at last, discern the model of different vehicle under the same vehicle brand in conjunction with the topological relation between each feature of headstock.
More domestic or external most of vehicle identification method mainly adopts license plate recognition technology to realize for the present invention and background technology, and discern types such as car, lorry from the contour of the vehicle contour feature, also do not find the brand that to discern vehicle, can also discern the information of the vehicle of dissimilar in the same brand (models) simultaneously.The method that the present invention relates to is directly to go out to discern vehicle brand and model in real time from the image that collects, and does not need to search type of vehicle information by existing database information, saved and database between comparison time and the interference of human factor; Compare with the Vehicle Type Recognition Technology of appearance profile and three-dimensional model, the method that the present invention relates to does not need to set up in advance the 3 d model library and the appearance profile storehouse of vehicle, but the headstock image of directly storing the different brands different model gets final product; Simultaneously recognition result of the present invention can combine with the Vehicle License Plate Recognition System of prior art etc., for the real-time follow-up of avoiding vehicles peccancy interline board etc. provides more reliable foundation.Through the trial run statistics, about about 95%～98%, the model discrimination of vehicle is greatly about about 92%～95% greatly for the brand recognition rate of vehicle.
Description of drawings
Fig. 1 is system's schematic block diagram of the said method of the present invention.
Fig. 2 is the said headstock localization method of a present invention schematic block diagram.
Fig. 3 is said each component flow diagram of orienting headstock from headstock of the present invention.
Fig. 4 is said vehicle brand of the present invention and model recognition principle process flow diagram.
Fig. 5 is the said method system handles of a present invention course of work schematic block diagram.
Embodiment
Find out by camera collection behind the vehicle front image by Fig. 1 Fig. 2, at first carry out the location of headstock, locate feature such as each parts such as car plate, car mark, air intake opening and headlight in the headstock then,, set up the topological relation between each parts according to the component feature of location; Then,, identify the brand of this vehicle in conjunction with the global feature information of headstock by the car mark is discerned; Under this brand,, identify the concrete model of vehicle under this brand at last according to the attribute of the topological relation between each feature and each feature itself.Because car is rigid body, and headstock has symmetry, adopted symmetry to come headstock is positioned, and mainly contains several symmetry: a gray scale symmetry; B edge feature symmetry, c direction symmetry; D profile symmetry.Utilize these several symmetrical standards that image is analyzed, can find the center line of vehicle, the profile of vehicle is expanded from alignment sideline, center, thereby obtains the width information of vehicle.Consider the influence of illumination condition, may a kind of symmetry can't accurately orient the headstock position, therefore can unite multiple symmetry method comes accurate in locating headstock position.Find out in headstock by Fig. 3, because the existence of each parts, cause the texture information of car face abundant, particularly there is tangible textural characteristics at parts places such as headlight, air intake opening and Che Biao, and a lot of detail textures are arranged all in each parts of headstock, in order effectively to locate each parts in the headstock, adopted a kind of multiple dimensioned local energy function to locate each parts of headstock.Essence is to select an optimal scale by calculating, under this yardstick, optimum radius by some candidate point place is to calculate energy function information in the border circular areas of s and the gray level skeleton function information give prominence to each parts in the headstock, thus effectively the cutting apart of realization each parts of headstock.Compare with traditional rim detection dividing method, this method has been utilized the gray scale and the texture information of image, and these information help to distinguish each parts of vehicle, because each parts of vehicle have certain contact on intensity and texture.Find out by Fig. 4 Fig. 5, discern according to the car target on the one hand, need component feature in addition on the one hand again in conjunction with headstock for the identification of vehicle brand.Car identifies other aspect: at first adopt template matches that the car mark is slightly discerned, the result to slightly identifying adopts a kind of yardstick invariant features transforming function transformation function further to discern then.No matter be to extract features such as angle point or edge, these features will change with rescaling, and are inaccurate to the characteristic matching of extracting on the different scale image, and yardstick invariant features transforming function transformation function has overcome this problem, and rotation and illumination are also had unchangeability.
Car head unit identification aspect: at first its coupling is discerned, then according to the further verification of the local feature of each parts from the global feature of headstock.By weighting scheme result's (car identifies other result and car head unit recognition result) of twice identification is carried out fusion treatment at last, finally identify the brand of vehicle.
The present invention has only utilized car mark and headstock information in the car body, and does not utilize whole locomotive body information.In identification during vehicle model, the different model of the car under the same brand is thought the combination at random of each car head unit, the combination at random of position and the incompatible consideration of random groups of geometric characteristic.Therefore pass through the parametric description of the feature of car head unit, likelihood function under the structure different scale, by merging the likelihood function under the different scale, obtain the likelihood function of this feature at last, discern the model of different vehicle under the same car mark in conjunction with the topological relation of each feature of headstock.
Fig. 5 has provided the course of work embodiment that whole recognition system of the present invention is handled.At first utilize camera acquisition vehicle front image sequence, according to the image of gathering, computer system adopts preprocessing means to handle to image earlier, merge the location that multiple symmetry realizes headstock then, in the headstock image of orienting, under yardstick S, calculate local energy function and gray level skeleton function and corresponding average energy function and average gray level skeleton function, whether the maximal value of judging absolute difference between energy function and the average energy function is greater than set threshold value (scope of threshold value is 5～8), if less than this threshold value, then change yardstick S, recomputate energy function and gray level skeleton function, if greater than this threshold value, then orientate this yardstick as best scale S, calculate energy function and gray level skeleton function under this yardstick then, realize cutting apart of each parts of headstock according to the function that is calculated, utilize characteristic parameter described to cutting apart each parts of headstock that obtain, after orienting headstock and each parts thereof, utilize the template matches criterion that car mark zone is mated earlier, the matching value of each class template is sorted, judge that whether the maximum match value is less than certain threshold value (establishing the scope 0.3～0.5 of threshold value), if less than this threshold value, need not this car mark is discerned, if greater than this threshold value, judge that then whether the maximum match value is greater than certain threshold value (establishing the scope 0.7～0.9 of threshold value), if greater than this threshold value, judge that whether absolute difference between the front two matching value is greater than certain threshold value (scope of establishing threshold value is 0.01～0.1), if greater than this threshold value, then the car of maximum match value correspondence is designated as car and identifies other result, if less than this threshold value, perhaps the maximum match value is less than setting threshold, then the extraction of yardstick invariant features is carried out in K maximum match zone before the choosing, identification car mark.Identify other result according to car, realize identification the vehicle brand in conjunction with the Global Information of headstock; After obtaining this vehicle brand result,, identify the vehicle model under this brand at last in conjunction with the feature of each parts of headstock and the topological relation between each parts.
Claims (1)
1, a kind of method of utilizing computer digital image treatment technology identification type of vehicle comprises image acquisition, image pre-service, image segmentation, image recognition, image output flow process, it is characterized in that following steps:
One) from the image of collection vehicle front end, orients headstock, utilize a kind of or merge multiple symmetry method image is analyzed, find out the center line of vehicle, the profile of vehicle is expanded from alignment sideline, center, obtain the width information of vehicle, cut apart identification and orient the headstock position; Described symmetry method can be profile symmetry, gray scale symmetry, edge symmetry or direction symmetry;
Two) from headstock, orient each parts of headstock,, utilize multiple dimensioned local energy function information and gray level skeleton function information to give prominence to each parts of headstock, thereby realize effectively cutting apart of each parts of headstock according to headstock each image of component gray scale and texture information; Described multiple dimensioned local energy function information and the gray level skeleton function information utilized is that input headstock image carries out carrying out rim detection after the image pre-service, is the center with certain marginal point, and radius is the regional calculating energy function E of S
sWith the gray level skeleton function T
sCalculate the average energy function E under the yardstick S
sWith average gray level skeleton function T
sWith energy function E
sWith average energy function E
sThe maximum value and the gray level skeleton function T of difference
sWith average gray profile function T
sThe maximum value and the set threshold of difference, greater than the T under the then calculating optimum yardstick S of threshold value
sAnd E
s, according to T
sAnd E
sCut apart and obtain each component feature of headstock, describe each component feature of headstock with parameter fitting then;
Three) in the car mark of orienting, utilize image recognition technology identification car mark, thereby obtain the vehicle brand, in car mark zone, carry out template matches, each class template maximum match value is sorted, with maximum match value and set threshold, if the maximum match value then need not to carry out the car sign not less than set threshold value, otherwise, judge that whether the maximum match value is greater than set threshold value, if the maximum match value is greater than set threshold value, whether the absolute value of judging front two maximum match value differences value again is greater than set threshold value, if the car of maximum match value correspondence is designated as recognition result, if not, the yardstick invariant features is carried out in front K maximum match zone extract, the car mark is discerned, if the maximum match value is less than set threshold value according to the feature of being extracted, the yardstick invariant features is carried out in front K maximum match zone extract, the car mark is discerned according to the feature of being extracted;
Four) in the image recognition flow process, identify the brand of vehicle according to the global feature information of car mark that obtains and headstock;
Five) utilize the feature of each parts of headstock and each feature between topological relation, in same brand, discern vehicle model, the parametric description of the feature by car head unit, likelihood function under the structure different scale, merge the likelihood function under the different scale, obtain the likelihood function of this feature at last, discern the model of different vehicle under the same vehicle brand in conjunction with the topological relation between each feature of headstock.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100225778A CN100485710C (en) | 2006-12-22 | 2006-12-22 | Method for recognizing vehicle type by digital picture processing technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100225778A CN100485710C (en) | 2006-12-22 | 2006-12-22 | Method for recognizing vehicle type by digital picture processing technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101196979A CN101196979A (en) | 2008-06-11 |
| CN100485710C true CN100485710C (en) | 2009-05-06 |
Family
ID=39547373
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100225778A Expired - Fee Related CN100485710C (en) | 2006-12-22 | 2006-12-22 | Method for recognizing vehicle type by digital picture processing technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100485710C (en) |
Families Citing this family (41)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101504717B (en) * | 2008-07-28 | 2012-07-11 | 上海高德威智能交通系统有限公司 | Characteristic area positioning method, car body color depth and color recognition method |
| CN101339601B (en) * | 2008-08-15 | 2011-09-28 | 张擎宇 | License plate Chinese character recognition method based on SIFT algorithm |
| TWI394095B (en) | 2008-10-22 | 2013-04-21 | Ind Tech Res Inst | Image detecting method and system thereof |
| CN102292700A (en) * | 2009-01-24 | 2011-12-21 | 惠普开发有限公司 | System and method for enhancing security printing |
| US8732172B2 (en) * | 2010-05-13 | 2014-05-20 | Institute Of Automation, Chinese Academy Of Sciences | Shape classification method based on the topological perceptual organization theory |
| CN102034093A (en) * | 2010-12-03 | 2011-04-27 | 天津工业大学 | Optimized calculation-based characteristic point matching method |
| CN102708385B (en) * | 2012-04-06 | 2014-07-16 | 张丛喆 | Method and system for comparison and recognition of three-dimensional vehicle types in video monitoring scenes |
| CN103065143A (en) * | 2012-12-30 | 2013-04-24 | 信帧电子技术（北京）有限公司 | Automobile logo identification method and device |
| CN103077397B (en) * | 2013-01-21 | 2016-02-24 | 信帧电子技术（北京）有限公司 | Vehicle-logo location method and vehicle-logo location system |
| CN103234530B (en) * | 2013-04-17 | 2015-04-22 | 江苏晓山信息产业股份有限公司 | City vehicle tire cleanliness detection method based on video signals |
| CN103473566B (en) * | 2013-08-27 | 2016-09-14 | 东莞中国科学院云计算产业技术创新与育成中心 | A kind of vehicle checking method based on multiple dimensioned model |
| CN103985230B (en) * | 2014-05-14 | 2016-06-01 | 深圳市大疆创新科技有限公司 | A kind of Notification Method based on image, device and notice system |
| CN104361343B (en) * | 2014-10-30 | 2018-08-31 | 中盟科技有限公司 | Vehicle type recognition method and its device |
| CN104392611B (en) * | 2014-11-17 | 2016-08-24 | 英华达（南京）科技有限公司 | A kind of identification high price automobile method and system |
| CN105095848B (en) * | 2014-12-01 | 2019-06-21 | 北京云识图信息技术有限公司 | The method and system of target identification |
| CN104778444B (en) * | 2015-03-10 | 2018-01-16 | 公安部交通管理科学研究所 | The appearance features analysis method of vehicle image under road scene |
| CN104966049B (en) * | 2015-06-01 | 2018-06-05 | 江苏航天大为科技股份有限公司 | Lorry detection method based on image |
| CN105184293B (en) * | 2015-08-29 | 2019-02-15 | 电子科技大学 | Vehicle-logo location method based on salient region detection |
| CN105160330B (en) * | 2015-09-17 | 2018-08-10 | 中国地质大学（武汉） | A kind of automobile logo identification method and vehicle-logo recognition system |
| CN106650550A (en) * | 2015-10-28 | 2017-05-10 | 中通服公众信息产业股份有限公司 | Vehicle model identification method and vehicle model identification system through image characteristics of vehicle mark and vehicle head |
| CN105448103B (en) * | 2015-12-24 | 2018-01-26 | 北京旷视科技有限公司 | Vehicle fake-license detection method and system |
| CN105512653A (en) * | 2016-02-03 | 2016-04-20 | 东南大学 | Method for detecting vehicle in urban traffic scene based on vehicle symmetry feature |
| CN105759683A (en) * | 2016-02-04 | 2016-07-13 | 李舒曼 | Wireless intelligent mobile phone charger |
| CN105629834B (en) * | 2016-02-04 | 2016-11-30 | 国网山东省电力公司费县供电公司 | Multi-functional charging system |
| CN108897055B (en) * | 2016-02-24 | 2020-02-21 | 北京君和信达科技有限公司 | Radiation source control method and quick-pass type security inspection system |
| CN107451511B (en) * | 2016-05-31 | 2020-12-15 | 优信拍（北京）信息科技有限公司 | System and method for extracting OBD data based on vehicle type information |
| CN108345820A (en) * | 2017-01-23 | 2018-07-31 | 许继集团有限公司 | High-tension apparatus image-recognizing method and device based on variety of components and component locations |
| CN108345896A (en) * | 2017-01-23 | 2018-07-31 | 许继集团有限公司 | High-tension apparatus image-recognizing method and device based on variety of components and scale of components |
| CN106971380A (en) * | 2017-03-13 | 2017-07-21 | 深圳市嘉和顺信息科技有限公司 | A kind of contrast enhancing and application of the visual saliency optimization method in golf course figure |
| CN106875693A (en) * | 2017-03-29 | 2017-06-20 | 广西信路威科技发展有限公司 | A kind of method and system of vehicle feature recognition |
| CN107256633B (en) * | 2017-05-09 | 2020-03-31 | 西安理工大学 | Vehicle type classification method based on monocular camera three-dimensional estimation |
| CN108038423B (en) * | 2017-11-22 | 2022-03-04 | 广东数相智能科技有限公司 | Automobile type identification method and device based on image identification |
| CN108090429B (en) * | 2017-12-08 | 2020-07-24 | 浙江捷尚视觉科技股份有限公司 | Vehicle type recognition method for graded front face bayonet |
| CN110769175B (en) | 2018-07-27 | 2022-08-09 | 华为技术有限公司 | Intelligent analysis system, method and device |
| CN110285870A (en) * | 2019-07-22 | 2019-09-27 | 深圳市卓城科技有限公司 | Vehicle spindle-type and wheel number determination method and its system |
| CN110598749A (en) * | 2019-08-13 | 2019-12-20 | 广东数鼎科技有限公司 | Image vehicle recognition implementation system based on convolutional neural network AI algorithm |
| CN110728174A (en) * | 2019-08-29 | 2020-01-24 | 恒大智慧科技有限公司 | Expansion gate control method and device based on vehicle identification and storage medium |
| CN110880003B (en) * | 2019-10-12 | 2023-01-17 | 中国第一汽车股份有限公司 | Image matching method and device, storage medium and automobile |
| CN111259885B (en) * | 2020-02-27 | 2023-11-24 | 深圳成子科技有限公司 | Wristwatch identification method and device |
| CN111942207A (en) * | 2020-08-14 | 2020-11-17 | 高丰 | Electric automobile intelligent charging management system based on big data |
| CN113124777B (en) * | 2021-04-20 | 2023-02-24 | 辽宁因泰立电子信息有限公司 | Vehicle size determination method, device and system and storage medium |
-
2006
- 2006-12-22 CN CNB2006100225778A patent/CN100485710C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN101196979A (en) | 2008-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100485710C (en) | Method for recognizing vehicle type by digital picture processing technology | |
| CN107729818B (en) | Multi-feature fusion vehicle re-identification method based on deep learning | |
| CN106056086B (en) | Vehicle brand type identifier method based on Fast Learning frame | |
| Silva et al. | An extended center-symmetric local binary pattern for background modeling and subtraction in videos | |
| CN108090429B (en) | Vehicle type recognition method for graded front face bayonet | |
| Zhang et al. | Three-dimensional deformable-model-based localization and recognition of road vehicles | |
| CN102156983B (en) | Pattern recognition and target tracking based method for detecting abnormal pedestrian positions | |
| CN105335702B (en) | A kind of bayonet model recognizing method based on statistical learning | |
| Li et al. | Vehicle detection based on the and–or graph for congested traffic conditions | |
| CN102663411B (en) | Recognition method for target human body | |
| CN105740886B (en) | A kind of automobile logo identification method based on machine learning | |
| CN102768726B (en) | Pedestrian detection method for preventing pedestrian collision | |
| CN108764096B (en) | Pedestrian re-identification system and method | |
| CN103049758A (en) | Method for realizing remote authentication by fusing gait flow images (GFI) and head and shoulder procrustes mean shapes (HS-PMS) | |
| CN101131728A (en) | Face shape matching method based on Shape Context | |
| CN103870808A (en) | Finger vein identification method | |
| CN114359876B (en) | Vehicle target identification method and storage medium | |
| CN103955496A (en) | Fast field tire trace pattern retrieval algorithm | |
| CN111435421A (en) | Traffic target-oriented vehicle weight identification method and device | |
| Wang et al. | V2I-CARLA: A novel dataset and a method for vehicle reidentification-based V2I environment | |
| CN104156697A (en) | Vehicle type recognition method under night bayonet scene | |
| Zheng et al. | OERFF: A vehicle Re-Identification method based on orientation estimation and regional feature fusion | |
| Lee et al. | independent object detection based on two-dimensional contours and three-dimensional sizes | |
| CN111862147A (en) | Method for tracking multiple vehicles and multiple human targets in video | |
| CN103559492A (en) | Car logo recognition device and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090506 Termination date: 20121222 |